The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves ingesting and analyzing untrusted data from the local codebase and external APIs (Microsoft Graph) to generate migration logic. This presents a surface for indirect prompt injection.
Ingestion points: The skill reads source code, environment variables (e.g., .env), application settings (e.g., appsettings.json), and XML custom policy files using grep and standard file reading tools. It also ingests data from Microsoft Graph API responses.
Boundary markers: There are no explicit delimiters or boundary instructions provided to help the agent distinguish between untrusted file content and its primary instructions during the analysis process.
Capability inventory: The skill enables the agent to execute shell commands (grep, npm, dotnet, pip), perform file system operations (read/write), and make network requests for discovery and documentation purposes.
Sanitization: The instructions do not specify any sanitization or validation logic for the content retrieved from external files or APIs before it is incorporated into the agent's context.
[EXTERNAL_DOWNLOADS]: The skill interacts with the Descope MCP Server (mcp.descope.com) and Microsoft Graph (graph.microsoft.com). These are legitimate interactions with the vendor's own infrastructure and well-known service providers. Additionally, the skill performs OIDC discovery by fetching configuration from dynamic issuer URLs.
[COMMAND_EXECUTION]: The skill provides the agent with complex, pre-defined shell command templates (primarily grep with specific exclusions) used to identify authentication patterns and legacy library usage within the local project directory. These commands are necessary for the discovery and triage phases of the migration.

azure-ad-b2c-to-descope